Classification of Alzheimer’s Progression Using fMRI Data

In the last three decades, development of functional magnetic resonance imaging (fMRI) has significantly contributed to understanding brain, brain mapping, and resting-state networks. Given recent successes deep learning in various fields, we propose a 3D-CNN-LSTM classification model diagnose health conditions with following classes: condition normal (CN), early mild cognitive impairment (EMCI), late (LMCI), Alzheimer’s disease (AD). The proposed method employs spatial temporal feature extractors, wherein former utilizes U-Net architecture extract features, latter long short-term memory (LSTM) features. Prior extraction, performed four-step pre-processing remove noise from fMRI data. comparative experiments, trained each models by adjusting time dimension. network exhibited an average accuracy 96.4% when using five-fold cross-validation. These results show that high potential for identifying progression analyzing 4D